gmane.comp.micro-kernel.l4.l4ka.general

RE: Is it possible to install L4 on Xen as a guest?

Chulmin Kim — 2012-05-10T08:15:28

Hi Jan, 

Any of those methods is OK for us. (Full-virt or Paravirt) 

We are planning to manage manycore machines using the xen+L4 combination.
Also, we will make some changes on each layer for the application specific
system. 


According to the xen wiki, 
It doesn't list L4 on the supported OS list for full-virtualization.
So, I asked the question for HVM(or full-virt) case. 


I didn't even expect the paravirtualization case. 
Is it also possible? Then, it will be more good for us. :)

Please, give me more information about this issue. 

Thanks!




to port
presented in

RE: Is it possible to install L4 on Xen as a guest?

Jan Stoess — 2012-05-10T07:43:29

Hello Chulmin,

May I ask what exactly do you want to achieve? Given Xen supports hardware virtualization, L4 as guest is supported out of the box. Or, do you want to port L4 to Xen as para-virtual guest, akin to the original Xen approach presented in the SOSP paper 2003, where they ported Linux? 

-Jan

--
Dr. Jan Stoess, KIT System Architecture Group

Is it possible to install L4 on Xen as a guest?

Chulmin Kim — 2012-05-10T02:17:23

Hi all, 

We are now interested in porint L4 on Xen for our research project. 

Does Any of you have an experience on this issue? 

Or any comment will be welcomed. 

Thanks. 


Sincerely,
Chulmin Kim

Re: Iterative calculations doesn't work - What could be the problem?

Håvard Ostnes — 2012-04-24T13:39:13

Thank you! This worked! I really appreciate it!

håvard

On 04/24/2012 02:54 PM, Peter Jin wrote:

Re: Iterative calculations doesn't work - What could be the problem?

Peter Jin — 2012-04-24T12:54:25

Probably because your iter_factorial call got optimized away. Try assigning
it to something, like

volatile unsigned f = iter_factorial(1000000001);

Peter


On Tue, Apr 24, 2012 at 7:50 AM, Håvard Ostnes <havard.ostnes< at >gmail.com>wrote:

Iterative calculations doesn't work - What could be the problem?

Håvard Ostnes — 2012-04-24T11:50:38

Hi!

I have the following code which successfully compiles and runs on L4 on Qemu and 
VMware.

The problem is that the calculations seems to finish immediately and looking at 
the CPU utilization it doesn't even seem to have performed the calculation at 
all.


-----CODE BEGINS-------
//LINUX
//#include <stdio.h>
//#include <unistd.h>
//L4
#include <l4io.h>
#include <l4/ipc.h> 
//#include <l4/types.h>

unsigned int iter_factorial(int n);
unsigned int recr_factorial(int n);

int main() {
  while(1)
    {
      printf("working\n");
      iter_factorial(1000000001);
      printf("Done\n");
      L4_Sleep(L4_TimePeriod(2000000));
      //sleep(2);
    }
}

unsigned int recr_factorial(int n) {
    return n>=1 ? n * recr_factorial(n-1) : 1;
}

unsigned int iter_factorial(int n) {
    int accu = 1;
    int i;
    for(i = 1; i <= n; i++) {
        accu *= i;
    }
    return accu;
}


-------- CODE ENDS----------


I really hope someone could take a look at this to help with my problem as I am 
stuck to the point w

RE: Can't set total quantum with L4_Set_Timeslice()

Jan Stoess — 2012-04-24T07:52:22

Hi Farid,

If you're referring to the user-space scheduling prototype described in the paper: the prototype has never made it into the mainline kernel. It was an experiment and had significant performance implications (as described in the paper), and we didn't want to pollute the mainline tree with it. If you're interested in the code let me know and I'll dig. 

If you're referring to the normal scheduling behavior in pistachio (RR with Prios): the preemption message has been specified in the manual, but has never been completely implemented (at least in L4Ka::Pistachio). Nobody has required or used that functionality, and since it requires careful implementation (see code comment snippet below) , it has been a "TODO" ever since. 

From kernel/src/api/v4/sched-rr/schedule.cc
/**
 * sends preemption IPC to the scheduler thread that the total quantum
 * has expired */
void rr_scheduler_t::total_quantum_expired(tcb_t * tcb)
{
    TRACEPOINT(TOTAL_QUANTUM_EXPIRED, "total quantum expired for %t\n", tcb);

Can't set total quantum with L4_Set_Timeslice()

Farid Hajji — 2012-04-23T15:09:28

Hello,

I'm experimenting with user space scheduling on L4Ka::Pistachio
(amd64) as specified in the paper
  "Towards Effective User-Controlled Scheduling for Microkernel-Based Systems"
and I'd like a worker thread to RPC its associated scheduler
thread, when it has used up its total quantum.

This however never seems to happen. In the KDB, I always see
that 'tq' is set to 0usec, in other words, to L4_Never. Is it right
that the user space scheduler would never be RPC-ed in this case?

So I've tried to change the 'tq' with a call to L4_Set_Timeslice().
But even then, the 'tq' always remain at 0usec, and what's worse,
the time slice also jumps to 0usec, even when set to some other
value.

Is the call to L4_Set_Timeslice() broken?

Attached is a simple test case, and the output of KDB before
and after the call to L4_Set_Timeslice().

Thanks in advance,
-Farid Hajji.
KickStart 0.12.56
Detected multiboot compliant loader
 kernel    (0x0010a000-0x0014ba45)   => 0x00b0e000
  (0x0010a1a0-0x001266a0) -> 0x00400000-0x

RE: C Library with L4 - Diet Libc library?

Jan Stoess — 2012-04-23T08:36:42

Hello Håvard,

Dietlibc should principally do, although not all functions may be supported without writing a proper backend for L4 akin to Linux backend. For instance, supporting pthreads on L4 through dietlibc will require you to map whatever dietlibc's backend for threads is onto L4 threads.

We had made use of dietlibc some time ago ourselves, for the afterburner project:
http://www.l4ka.org/projects/virtualization
but support was limited to those functions that do without a backend (eg. math functions).


Side note, if you're just looking for a malloc implementation, you may want to consider dlmalloc
ftp://g.oswego.edu/pub/misc/malloc.c 

Best,
-Jan


--
Dr. Jan Stoess, KIT System Architecture Group
Phone: +49 (721) 6084 4056 Cell: / +49 (178) 149 5341

C Library with L4 - Diet Libc library?

Håvard Ostnes — 2012-04-18T13:11:42

Hi!

I have googled around and saw someone mentioning that the 
diet libc library could be used with L4 Pistachio.

I was wondering if someone knows how to successfully compile
an L4 application using the diet libc library. 
(or another similar C library).

The reason is I need to use malloc and other functions in my application.

Best regards,
håvard ostnes

Re: Hello World - No Output

Håvard Ostnes — 2012-03-07T00:09:37

"hello

I appreciated your help. I had to edit configure, configure.in and Makefile.in 
with pointers to where my hello world application was located.

This is my hello world roottask:
hello.cc:
------------------------------------------------------------------
#include <l4io.h>
#include <l4/ipc.h>


int main (void)
{
  int i=0;
   L4_Time_t timeout;
   timeout = L4_TimePeriod(3000000);

   printf("Starting infinite loop\n");

   while(true){
      printf ("Sleeping. Count %i \n",i++);
      //sleeping for 10 seconds
      L4_Sleep (timeout);
   };

   return 0;
}

-------------------------------------------------------------------

I have one more question regarding CPU usage. Even though this roottask only 
prints one line of text every three seconds the KVM virtual machine constantly 
demands between 1 and 5% of the CPU.

The screenshot below show the CPU usage in percentage in column number 9:

Screenshot_at_2012-03-07_01:02:40.png:
[URL=http://imageupload.org/en/file/195350/screenshot-at-2012-03-07-01-

Re: Hello World - No Output

Semion Prihodko — 2012-02-16T14:34:22

You need to redirect serial port to console (your printf outputs into it).

2012/2/16 Håvard Ostnes <havard.ostnes< at >gmail.com>

Hello World - No Output

Håvard Ostnes — 2012-02-16T12:05:15

Hi!

I have spent the past two days on learning how to make a simple roottask "hello
world" application.

I can successfully compile the following code <hello.cc> file:

#include <l4io.h>

int main (void)
{
   
   bool hello = true;
   
   if (hello)
     {
printf ("\nPlease select ipc type:\n");
return 0;
     }

}

My Makefile looks like this:
srcdir=        < at >srcdir< at >
top_srcdir=    < at >top_srcdir< at >
top_builddir=    < at >top_builddir< at >

include $(top_srcdir)/Mk/l4.base.mk

PROGRAM=    helloworld
PROGRAM_DEPS=    $(top_builddir)/lib/l4/libl4.a \
 $(top_builddir)/lib/io/libio.a

SRCS=        hello.cc

LIBS+=        -ll4 -lio
LDFLAGS+=    -Ttext=$(ROOTTASK_LINKBASE)
CPPFLAGS+=    -I$(srcdir)/$(ARCH)

include $(top_srcdir)/Mk/l4.prog.mk


my menu.lst looks like this:

root (fd0)
default=0
timeout=3

title L4Ka::Pistachio
 kernel /kickstart
 module /x86-kernel
 module /sigma0
 module /helloworld
 

I guess there might be a problem somewhere, especially my hunch is on the
Makefile? (I am not a programmer by professio

RE: L4La-Pistachio: using idl4 for amd64

Jan Stoess — 2012-02-10T07:54:27

You can always produce generic stub code (written in c), which will work for every architecture, including amd64. The platform switches are only for optimizations. Have a look at the IDL4 website and manual:
http://www.l4ka.org/56.php

-Jan


--
Dr. Jan Stoess, KIT System Architecture Group
Phone: +49 (721) 6084 4056 Cell: +49 (1520) 160 1474

From: l4ka-bounces< at >ira.uni-karlsruhe.de [mailto:l4ka-bounces< at >ira.uni-karlsruhe.de] On Behalf Of Semion Prihodko
Sent: Thursday, February 09, 2012 10:01 PM
To: l4ka< at >ira.uni-karlsruhe.de
Subject: L4La-Pistachio: using idl4 for amd64

Hi, guys.

Have a little question. As I understand, the Idl4 stub generator does not directly support amd64 (only ia32/arm). Am I right? In case I'm right, is there a way to adopt the tool to this architecture? Thanks.

Fwd: cannot boot the compiled pistachio iso

rukshan rajapaksha — 2012-02-10T07:49:13

---------- Forwarded message ----------
From: rukshan rajapaksha <ruqee8< at >gmail.com>
Date: Thu, Feb 9, 2012 at 11:08 AM
Subject: Re: cannot boot the compiled pistachio iso
To: Jens Kehne <jens< at >int80.de>


hi guys,

    Finally i compiled the kernel and run perfectly.thank you very much for
the help.i tried with simple helloworld app(i replace pingpon.cc with my
helloworld .cc and compile pistachio again) it also worked perfectly.Now i
want to know how this pistachio works i mean from the booting.can you send
me a hierarchical description of pistachio(something like a graph).

as example   initial programme--->kernel------->------>
like that

Could you please tell me where this os start(a file or a code)

** my english is not that much good sorry for that***

thank you
rukshan

L4La-Pistachio: using idl4 for amd64

Semion Prihodko — 2012-02-09T21:00:34

Hi, guys.

Have a little question. As I understand, the Idl4 stub generator does not
directly support amd64 (only ia32/arm). Am I right? In case I'm right, is
there a way to adopt the tool to this architecture? Thanks.

16MB memory limit?

Håvard Ostnes — 2012-02-08T18:17:45

greetings!

I have tried running pistachio with maxmem < 16M but it won't allow me to.
I'm wondering if it might be possible to execute the kernel with  maxmem= < 16M
of memory. And also, how to achieve this. 

The kernel itself can run with only 1M of RAM, which leaves 15M of RAM which I
want to reduce to the smallest amount possible.

Is it possible to reduce it significantly and how feasible is it do achieve it
within a short amount of time?


cheers,
håvard

Re: How to create a working build environment

Håvard Ostnes — 2012-02-01T14:46:21

using a recent compiler, it
ia32 on Ubuntu oneiric. 

I discovered that the problems was related to Virtualbox, not GCC/G++.
I installed the current build-essential package onto my Debian Lenny box and the
kernel now runs smoothly.

For those interested, I have made a how-to on bulding L4 kernel and configuring
it to boot using GRUB2.

http://dl.dropbox.com/u/11080637/How%20To%20Build%20L4.pdf

thank you for your reply!

håvard

Re: Oops! No BIOS memory Map

Håvard Ostnes — 2012-02-01T14:43:02


hi!
I switched to VMWare and it worked like magic. The kernel runs perfectly with
the GCC version currently included in the build-essential package.

I have another question though; I wish to start adding things like a shell (a
BASH alternative perhaps?) and file systems etc.

Do you have any pointers as where I can start reading about adding a shell and
other modules to the kernel?

håvard

RE: How to create a working build environment

Jan Stoess — 2012-01-30T20:05:09

Hi Håvard Ostnes,

Is there any reason why you're trying 3.2? You'll most probably be easier off using a recent compiler, it says 3.2 or higher. 4-versions generally work; I'm currently using 4.6 for ia32 on Ubuntu oneiric. 

-Jan

--
Dr. Jan Stoess, KIT System Architecture Group
Phone: +49 (721) 6084 4056 Cell: +49 (1520) 160 1474

Re: Oops! No BIOS memory Map

Jens Kehne — 2012-01-30T19:57:46

Am 30.01.2012 19:24, schrieb Håvard Ostnes:

Hi Håvard,


I can't seem to find a previous post from you. I don't know what
happened there, maybe my spam filter caught it.


It's quite possible that it's Virtualbox. I just tried and the kernel
apparently gets stuck. Alternatively, you can try a different
virtualization software. KVM works for me.

Hope I could help,

Jens